Eventually, this will be a set of functions specifically for single nucleotide polymorphisms (SNPs), which are biallelic markers. This is particularly relevant to the genomewide association studies (GWAS) using GeneChips and in line with the classic generalised single-locus model. snp.HWE is from Abecasis's website and yet to be adapted for chromosome X.
snp.ES(beta, SE, N)snp.HWE(g)
PARn(p, RRlist)
snp.PAR(RR, MAF, unit = 2)
Regression coefficient.
Standard error for beta.
Sample size.
Observed genotype vector.
genotype frequencies.
A list of RRs.
Relative risk.
Minar allele frequency.
Unit to exponentiate for homozygote.
Jing Hua Zhao, Shengxu Li
snp.ES provides effect size estimates based on the linear regression coefficient and standard error. For logistic regression, we can have similar idea for log(OR) and log(SE(OR)).
snp.HWE gives an exact Hardy-Weinberg Equilibrium (HWE) test and it return -1 in the case of misspecification of genotype counts.
snp.PAR calculates the the population attributable risk (PAR) for a particular SNP. Internally, it calls for an internal function PARn, given a set of frequencies and associate relative risks (RR). Other 2x2 table statistics familiar to epidemiologists can be added when necessary.